Terms & Privacy
Version 0.1 (draft) · October 27, 2026 · plain-language summary first, full terms to follow after review.
The short version
- It's free because what you upload and correct helps us improve our speech models.
- Your data is private. Your audio, text and corrections are visible only to you unless you turn on a share link.
- We use it only to improve Devniks models. We never sell it, never publish it, and never give it to advertisers.
- You can delete it. Any item, or your whole account, at any time.
1. Who we are
Devniks Voice (voice.devniks.com) is a service of Devniks. "We" means Devniks. Contact: [email protected] (placeholder address).
2. Using the service
- You must be at least 13 years old. Children under 13 may not create an account.
- One account per person. Keep your password and API keys secret.
- Only upload audio you have the right to use. Don't upload recordings of people who didn't agree to be recorded where the law requires consent.
- No illegal, abusive or deceptive use, including faking someone's voice.
- The free service comes with daily limits and no uptime guarantee. We may change limits or features.
3. What we store
- Your account: name, email, sign-in method (Google or email), and when you accepted these terms.
- What you upload or type: audio files, text for speech, and the results.
- What you change: transcript corrections (with the original model output kept next to them), TTS ratings and comments.
- Basic usage logs: dates, file length, model version, IP address for security and abuse prevention.
4. How we use it
Your audio, text and corrections are stored privately and used by Devniks only to provide the service and to improve and retrain our speech models. Some samples may be listened to by a small number of Devniks reviewers bound by confidentiality, for example to spot-check corrections. We never sell or publish your data, and we never use it for advertising.
Models we train may be offered by Devniks to others. Those models do not contain your files, and we take care that they don't repeat private content word for word.
5. Your voice (separate consent)
Advanced ASR
Advanced ASR uses Google's Gemini API to recheck audio and its draft transcript. It can run automatically for low-confidence results when the upload option is enabled, or when you request a second pass. We keep the original output and the separate Advanced ASR result. This processing is separate from Sixday searches; Voice recordings are not added to Sixday search history. Google's applicable API data terms also apply. You can turn off automatic Advanced ASR before uploading.
- Using your voice recordings to train voice (text-to-speech) models needs a separate opt-in. You can turn it on or off in Settings at any time.
- Voice cloning: only clone your own voice or a voice you have clear permission to use. Cloned voices stay private to your account. We may remove clones that break this rule.
6. The API
Everything sent through the API is covered by these same terms.
7. Your choices
- Download a copy of your data from Settings.
- Delete any item, or delete all your data and your account. Deleted data is removed from our storage and from future training sets within 30 days; models already trained before deletion are not retrained to remove it.
- Withdraw voice-training consent at any time; it applies going forward.
8. Security & storage
Data is encrypted in transit and at rest, stored with our hosting providers, and access is limited to people who need it to run or improve the service.
9. Changes
If we change these terms in an important way we'll tell you by email and ask you to accept the new version before you keep using the service.
Questions for the lawyer: data location (US/EU), retention period for logs, GDPR / CCPA rights wording, whether the 30-day deletion window and "models not retrained" sentence are acceptable, minimum age (13 vs 16 for EU users).